Add utf8::read_codepoint that both gets the codepoint and advance iterator

2 files changed, 17 insertions, 11 deletions

diff --git a/src/highlighters.cc b/src/highlighters.cc
index 052ff657..9c818a89 100644
--- a/src/highlighters.cc
+++ b/src/highlighters.cc
@@ -861,20 +861,19 @@ void expand_unprintable(const Context& context, HighlightFlags flags, DisplayBuf
                 for (auto it  = buffer.iterator_at(atom_it->begin()),
                           end = buffer.iterator_at(atom_it->end()); it < end;)
                 {
-                    Codepoint cp = utf8::codepoint<utf8::InvalidPolicy::Pass>(it, end);
-                    auto next = utf8::next(it, end);
+                    auto coord = it.coord();
+                    Codepoint cp = utf8::read_codepoint<utf8::InvalidPolicy::Pass>(it, end);
                     if (cp != '\n' and not iswprint(cp))
                     {
-                        if (it.coord() != atom_it->begin())
-                            atom_it = ++line.split(atom_it, it.coord());
-                        if (next.coord() < atom_it->end())
-                            atom_it = line.split(atom_it, next.coord());
+                        if (coord != atom_it->begin())
+                            atom_it = ++line.split(atom_it, coord);
+                        if (it.coord() < atom_it->end())
+                            atom_it = line.split(atom_it, it.coord());
 
                         atom_it->replace(format("U+{}", hex(cp)));
                         atom_it->face = { Color::Red, Color::Black };
                         break;
                     }
-                    it = next;
                 }
             }
         }
diff --git a/src/utf8.hh b/src/utf8.hh
index 545255fd..09d66b99 100644
--- a/src/utf8.hh
+++ b/src/utf8.hh
@@ -110,7 +110,7 @@ struct Pass
 // is pointed by it
 template<typename InvalidPolicy = utf8::InvalidPolicy::Pass,
          typename Iterator>
-Codepoint codepoint(Iterator it, const Iterator& end)
+Codepoint read_codepoint(Iterator& it, const Iterator& end)
 {
     if (it == end)
         return InvalidPolicy{}(-1);
@@ -124,14 +124,14 @@ Codepoint codepoint(Iterator it, const Iterator& end)
         return InvalidPolicy{}(byte);
 
     if ((byte & 0xE0) == 0xC0) // 110xxxxx
-        return ((byte & 0x1F) << 6) | (*it & 0x3F);
+        return ((byte & 0x1F) << 6) | (*it++ & 0x3F);
 
     if ((byte & 0xF0) == 0xE0) // 1110xxxx
     {
         Codepoint cp = ((byte & 0x0F) << 12) | ((*it++ & 0x3F) << 6);
         if (it == end)
             return InvalidPolicy{}(cp);
-        return cp | (*it & 0x3F);
+        return cp | (*it++ & 0x3F);
     }
 
     if ((byte & 0xF8) == 0xF0) // 11110xxx
@@ -142,11 +142,18 @@ Codepoint codepoint(Iterator it, const Iterator& end)
         cp |= (*it++ & 0x3F) << 6;
         if (it == end)
             return InvalidPolicy{}(cp);
-        return cp | (*it & 0x3F);
+        return cp | (*it++ & 0x3F);
     }
     return InvalidPolicy{}(byte);
 }
 
+template<typename InvalidPolicy = utf8::InvalidPolicy::Pass,
+         typename Iterator>
+Codepoint codepoint(Iterator it, const Iterator& end)
+{
+    return read_codepoint(it, end);
+}
+
 template<typename InvalidPolicy = utf8::InvalidPolicy::Pass>
 ByteCount codepoint_size(char byte)
 {